This article is going to look into the contributions of each country to the ocean's plastic pollution, based on a study perform by the Ocean Clean-Up, a charity dedicated to the clean-up of the great pacific garbage patch and all ocean plastics. The study was published in the journal Science Advances. The results are also displayed in a great interactive world map on the Ocean Clean-Up website. The study attempts to model the plastic pollution contribution of every river in the world, and outputs the rivers estimated to have non-neglible plastic pollution. The modelling was based on a data collected from around 130 rivers. The model was applied to 100,000 rivers, with around 30,000 rivers estimated to produce plastic pollution. The exact modelling process can be viewed on the Science Advances website, linked above.
The purpose of this article was purely out of interest to see what I could gleam from this data. It should be noted that any conclusions drawn are based on the modelled data and hence the exact amounts of plastic pollution for each country is unknown and the confidence intervals given for the levels of pollution in the study are fairly wide.
The data I have used comes from the article mentioned above and is hosted on figshare and the country data that comes from Kaggle. The data output from the study comes in the form of a GIS shapefile, with data on the 30,000 rivers. I used a reverse-geocoder to get this by country and joined it to the Kaggle dataset. Let's have a quick look at a snippet of the data:
| Alpha-3 code | Country | Alpha-2 code | Region | Population | Area (sq. mi.) | Pop. Density (per sq. mi.) | Coastline (coast/area ratio) | Net migration | Infant mortality (per 1000 births) | GDP ($ per capita) | Literacy (%) | Phones (per 1000) | Arable (%) | Crops (%) | Other (%) | Climate | Birthrate | Deathrate | Agriculture | Industry | Service | plastic poll | no. rivers | Coastline | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | AGO | Angola | AO | SUB-SAHARAN AFRICA | 12127071.0 | 1246700.0 | 9.7 | 0.13 | 0.00 | 191.19 | 1900.0 | 42.0 | 7.8 | 2.41 | 0.24 | 97.35 | NaN | 45.11 | 24.20 | 0.096 | 0.658 | 0.246 | 868.300774 | 77 | 162071.00 |
| 1 | ALB | Albania | AL | EASTERN EUROPE | 3581655.0 | 28748.0 | 124.6 | 1.26 | -4.93 | 21.52 | 4500.0 | 86.5 | 71.2 | 21.09 | 4.42 | 74.49 | 3.0 | 15.11 | 5.22 | 0.232 | 0.188 | 0.579 | 1565.018726 | 36 | 36222.48 |
| 2 | ARE | United Arab Emirates (the) | AE | NEAR EAST | 2602713.0 | 82880.0 | 31.4 | 1.59 | 1.03 | 14.51 | 23200.0 | 77.9 | 475.3 | 0.60 | 2.25 | 97.15 | 1.0 | 18.96 | 4.40 | 0.040 | 0.585 | 0.375 | 14.021303 | 8 | 131779.20 |
| 3 | ARG | Argentina | AR | LATIN AMER. & CARIB | 39921833.0 | 2766890.0 | 14.4 | 0.18 | 0.61 | 15.18 | 11200.0 | 97.1 | 220.4 | 12.31 | 0.48 | 87.21 | 3.0 | 16.73 | 7.55 | 0.095 | 0.358 | 0.547 | 4016.384342 | 117 | 498040.20 |
| 4 | ATG | Antigua and Barbuda | AG | LATIN AMER. & CARIB | 69108.0 | 443.0 | 156.0 | 34.54 | -6.15 | 19.46 | 11000.0 | 89.0 | 549.9 | 18.18 | 4.55 | 77.27 | 2.0 | 16.93 | 5.37 | 0.038 | 0.220 | 0.743 | 1.802356 | 4 | 15301.22 |
To start, we'll just look at the plastic pollution by country:
As we can see South-East Asia is by far the worst offender when it comes to river palstic pollution, with the Phillipines having the highest at 356,000 MT/year, almost 3 times the second country India. The mapcan be used to interactively explore different areas, with the hover labels showing the plastic pollution levels.
To get a better idea of proportion of the global plastic pollution that the most polluting countries contribute, lets look at the percentage of total plastic pollution:
The plot shows the skew of the river pollution to a small number of highly polluting countries, with the top 10 countries countributing 82.6% of the yearly polution. Note the third largest sector is the other countries not included in the legend.
It is also interesting to see this split by region. The graph shows the top 30 most polluting countries sorted by their region. Hover on the bars to see which countries are included.
As imagined Asia (Ex-Near East) has by far the greatest impact.
Next we will look at the plastic pollution of the countries with respect to their populations, area and number of rivers in the dataset.
We can see that some of the Asian countries still feature strongly, with the Phillipines having the highest pollution per capita, however, countries like China and India don't feature due to their huge populations. We also see small countries in South America and the Carribean having very significant plastic pollution per capita.
We see similar trends when looking at the pollution per country area, with the Phillipines and the small Carribean islands featuring heavily again.
When thinking about which countries to target it might make sense to tackle the countries with the most polluting rivers. Here our results take a turn as we see the African countries having the highest pollution per river, although it is a much more even spread in the regrad with Benin being somewhat of an outlier. In this data I have used the number of rivers as the 30,000 rivers output in the data, and hence non-polluting rivers are not included here.
In order to get a better idea of the most polluting rivers we need to look at the data by inidividual river.
The graph again shows that many of the most polluting rivers are in the Phillipines, including the largest by far - a river in Manila - that produces 62.6k metric tons per year.
So this shows us which rivers need to be targetted most, however, what would the impact on the total plastic emissions be if we targetted the most polluting rivers?
The figure below shows the cumulative pollution as a percentage of total pollution by the 5000 most polluting rivers in the world. As you can see 80% of the pollution comes from the 1650 most polluting rivers. This was one of the main findings of the study.
Next, I wanted to explore relationships a country, its features and the amount of plastic pollution. Let's look at correlations between plastic pollution and the features in our data.
The largest positive correlations are for plastic pollution are with the number of rivers, the population, the coastline and the % of land used for crops. We need to take the no. rivers result with a heacy pinch of salt, since our dataset only inlcudes the polluting rivers and so the correlation is an obvious one. The coastline and population are also expected to be positively correlated with plastic pollution. I was surprised however that the coast/area ratio was not correlated to plastic pollution, since it would imply populations live closer to the coast.
The largest negative correlations are from GDP per capita, phones per person and % of land not used in farming. This would imply that richer countries produce lower plastic waste. What makes this interesting to me is that it implies it is not the level of production in the country (which would in turn mean higher volumes of plastic) it is the management of that waste that makes a difference.
It should be noted that not of these correlations are particularly strong and so many other factors come into play. Next we will look at what the study included in the model to predict plastic pollution.
The study's model uses the following features in their model:
| Area | Coastline | Rainfall | Probability | Mismanaged Waste | River Pollution | RP/MW | C/A | CxP/A | |
|---|---|---|---|---|---|---|---|---|---|
| Country | |||||||||
| Albania | 28486 | 362 | 1117 | 0.0156 | 69833 | 1565 | 0.0224 | 0.0100 | 14 |
| Algeria | 2316559 | 998 | 80 | 0.0009 | 764578 | 5774 | 0.0076 | 0.0004 | 0 |
| Angola | 1247357 | 1600 | 1025 | 0.0009 | 236946 | 860 | 0.0036 | 0.0010 | 1 |
| Antigua and Barbuda | 443 | 153 | 996 | 0.0308 | 627 | 2 | 0.0029 | 0.3000 | 344 |
| Argentina | 2779705 | 4989 | 567 | 0.0026 | 465808 | 4137 | 0.0089 | 0.0020 | 1 |
Probability is the probability that mismanaged waste will make it to the ocean, and is based on the layout of the river basin, for example the gradient of land and land use and other factors like the amount of rain. RP/MW is the river pollution over the mismanaged waste, C/A is coastline over area and CxP/A is coastline x probability over area.
First, we will again look at the correlations between the data, using a heatmap:
As expected the highest correlation to river pollution is the mismanaged waste. Which contributes much more than the probability of the waste making it to the ocean. This might suggest waste management programs might be more impactful than attempting to stop the plastic form entering the rivers and therefore the oceans.
When looking at the probability of mismanaged plastic reaching the rivers, it is strongly correlated with rainfull, but not correlated with the amount of mismanaged waste, nor is it strongly correlated to the amount coastline or area of the country. Note this is an average of the probabilities of each grid of each river basin in the country, and hence is not equal to the RP/MW.
To illustrate this the graph below shows mismanaged waste against the probability of plastic reaching the ocean, with the overall plastic pollution overlaid with a colour gradient and marker size.
From this we can clearly see the larger circles generally in the lower range of the overall probabilities implying the mismanaged waste is the greater issue compared to the probability of plastic reaching the ocean.
We note the countries with very high probabilities are small countries, mostly islands, which have very low plastic river pollution. This is expected as the plastic has less distance to travel to reach the sea but conversely there is very low amounts of plastic mismanaged. Note countries like India and China have relatively low probabilities but very large amounts of mismanaged waste.The Phillipines seems to be the perfect storm of both.
Finally, looking at the impacts on the probability of plastic reaching the ocean, we can see the rainfall has the highest positive correlation. We can look at the proportion of mismanaged waste that reaches the ocean compared to the rainfall.
It is clear to see the links between the proportion of plastic that reaches the ocean and rainfall. Unfortunately we do not have data on the other variables that impact the probability, such as the gradient of the river basin, as a result I will leave the investigation here.
Overall, I find the study very interesting and highlights the scale of the problem but also gives a great platform to show how it can be dealt with, and the Ocean Clean-Up is well on its way to tackling some of the most polluting rivers in the world. Keep it up!